Abstarct: Scouring of bridge piers is one of the main causes of bridge failure. Therefore, accurate prediction of scour depth around bridge piers is important both for determining sufficient depth for foundation of new bridge piers and for evaluating and monitoring existing bridge safety. In this study, a new artificial intelligence (AI) model, called Intelligent Inference Radial baxses Neural Network (IFRIM), is presented. This model combines radial neural network (RBFNN), fuzzy logic (FL) and optimization algorithms, in which the input data using fuzzy logic is converted to fuzzy then the most optimal fuzzy network parameters. The resulting nerve is obtained by the optimization algorithm and finally the fuzzy outputs obtained by defuzzification functions are converted to numerical outputs. In the lead study, bee colony optimization (ABC), genetic algorithm (GA) and population distribution (PSO) algorithms were used on two laboratory data sets and field data collected by the US Mapping Agency. Finally, the verification is verified by a series of other field data collected by Froholich. In this study, root mean square error (RMSE), mean absolute error (MAE) and correlation coefficient (CC) between predicted values and actual values were used to describe the analyzes. These values for the bee colony optimization algorithm are respectively 0.1, 0.019 and 0.81 for laboratory data, 0.089, 0.018 and 0.93 for field data, and 0.085, 0.019 and 0.94 for Froholich data other performance algorithms, respectively, compared to other algorithms in total performance. Better. Comparing the performance of this study with other similar studies on this topic, it can be concluded that the error in this study is reduced, which is discussed in detail in the context of this study.